Tuesday, 2017-12-05

*** tpb has joined #timvideos00:00
*** samsagaz has quit IRC00:12
*** samsagaz has joined #timvideos03:24
*** rohitksingh_work has joined #timvideos03:35
*** rohitksingh_work has quit IRC03:38
*** CarlFK has joined #timvideos03:42
*** ChanServ sets mode: +v CarlFK03:42
*** CarlFK has quit IRC03:43
*** rohitksingh has joined #timvideos13:14
*** sb0 has joined #timvideos13:36
shorneshenki: I had a go at running buildroot with musl14:07
shorneand no issues running ping14:07
*** springermac has quit IRC14:53
*** springermac has joined #timvideos14:53
*** CarlFK has joined #timvideos16:34
*** ChanServ sets mode: +v CarlFK16:34
*** sb0 has quit IRC16:49
*** FelixVi has joined #timvideos17:33
FelixVihi, has anybody here setup the environment in cygwin before?17:34
rohitksinghFelixVi: cr1901_modern would be the best bet for Cygwin. He has used it quite much.17:37
FelixVirohitksingh: Thanks! I'll see if he wants to chime in a little bit later17:39
rohitksinghFelixVi: Anytime! :)17:39
CarlFKyou cant use native windows?  (I have no clue)17:40
*** rohitksingh has quit IRC17:41
FelixViCarlFK: I am not sure actually - but since the default dev instructions are for ubuntu, cygwin could be closer17:45
FelixViDon't judge me for the cygwin thing - but work forces me to do most things in Windows17:45
CarlFKdo you do other fpga stuff in any os?17:46
FelixViYeah, I do most work with ISE and some with Vivado - all Windows, even though often based on scripts17:47
FelixViSo on that end I should be OK17:47
CarlFKI would think it is easier to "do that" than "use cygwin"17:47
CarlFKbut I don't find any of it easy... so ...17:48
CarlFKhave you tried Win 10 Linux subsystem?17:48
FelixViNo, I am currently on Win717:49
CarlFKI have seen people try it and it works.   so maybe that's better than cygwin17:49
FelixViif it's worth upgrading, I might be able to do that17:49
FelixViyeah, thanks! Thus far, Win7 has been a known evil17:50
CarlFKI've seen people apt install gimp firefox and they work.17:58
FelixViThis seems doable, too18:00
FelixVijust a couple issues with path - the env script doesn't run, but I'm trying to do it manually18:00
FelixViis there a way to just make a basesoc image (default make tries to use hdmi2usb)18:15
mithroFelixVi: why do you continue to try and do things in Windows?18:16
FelixVibecause I am on Windows for work18:16
FelixViI have to so I can use other software I have to have accesss to18:17
FelixVimithro: my goal is to have a LM32 toolchain that people can use from Windows machines18:25
mithroFelixVi: So, that is something I would like to support18:25
mithroFelixVi: So, the best way to do that is to start with conda for Windows18:26
mithroFelixVi: I tried to get our conda packages here to compile under windows here -> https://github.com/timvideos/conda-hdmi2usb-packages/tree/windows18:26
tpbTitle: GitHub - timvideos/conda-hdmi2usb-packages at windows (at github.com)18:26
FelixViIs that because of how they have packages arranged?18:26
FelixViI am currently using vanilla python with virtualenvwrapper18:27
mithroFelixVi: Yeah, we use conda for binary delivery18:27
FelixViCool, I think I should be able to use conda without breaking anything else I got going18:27
mithroFelixVi: If your in the Scientific realm, then you probably already using conda for other things?18:27
mithroFelixVi: yeah conda gives you a nice self contained environment18:28
FelixVias of now I don't - virtualenvwrapper is all I'm using18:28
FelixViso I just create virtual environments manually18:28
FelixVibut the concept is the same - seperate environments for different projects18:29
mithroConda includes all the things like numpy, scipy, pandas that scientists tend to use18:29
FelixViLet me install conda and check it out18:31
FelixViIn my environments, I install all packages myself as I need specific versions for different things18:31
mithroFelixVi: Take a look at this -> https://github.com/timvideos/conda-hdmi2usb-packages/blob/windows/.appveyor.yml18:31
tpbTitle: conda-hdmi2usb-packages/.appveyor.yml at windows · timvideos/conda-hdmi2usb-packages · GitHub (at github.com)18:31
FelixViI assume you can install packages via setuptools in conda even without a conda package available, right?18:31
mithroFelixVi: yes18:32
FelixVialright, give me a sec here18:34
FelixViappveyor is something I'm not familiar with18:35
FelixViso, you are already registered and that's how you want to deliver the environment?18:36
FelixVimithro: Could you give me a few pointers about how I should install the toolchain in Windows?18:42
FelixViI'll go through and take notes so it can be turned into setup instructions easily18:43
mithroappveyor is basically a windows CI environment - if we can get that working then we will have windows conda packages18:47
FelixViis there already a registration?18:47
FelixVisorry, I am a bit lost about how to get started18:48
mithroFelixVi: Install conda, checkout that branch of our github repo and see if you can "conda build XXXX" the packages on Windows18:49
FelixVialright - any conda version?18:49
FelixVithere's miniconda and anaconda18:50
mithroFelixVi: Either will work but miniconda is smaller :-P18:50
FelixVicool, I'm on it18:50
mithroFelixVi: I actually have finally gotten access to Windows machine (as of last week) which I can also do testing like this on19:10
FelixViwell, I hope I can do most of the legwork for you19:11
FelixViBut if you can help figure out some of the issues, that'd be great19:11
mithrocr1901_modern: poke! see -^19:14
FelixVimithro: This is the error I get from flterm: https://github.com/FelixVi/HDMI2USB-litex-firmware/blob/master/win_conda.md19:16
tpbTitle: HDMI2USB-litex-firmware/win_conda.md at master · FelixVi/HDMI2USB-litex-firmware · GitHub (at github.com)19:16
mithroFelixVi: I'd start with binutils-lm32-elf19:17
FelixVimithro: does conda come with make or do I need msys or cygwin for that?19:21
mithroFelixVi: You'll need MSVC++ I think?19:21
mithroFelixVi: I'm actually unsure what I was using on AppVeyor -- they seem to have multiple compilers installed -> https://www.appveyor.com/docs/build-environment/#pre-installed-software19:22
tpbTitle: Build Environment | AppVeyorAppVeyor Logo (at www.appveyor.com)19:22
FelixVithat makes sense19:23
FelixViI'm missing isl >= 0.1519:23
mithro"conda install isl" doesn't work?19:23
FelixViit says it's missing in current channels19:24
mithroFelixVi: Oh - it looks like I only was working on openocd  https://github.com/timvideos/conda-hdmi2usb-packages/tree/windows/openocd19:26
tpbTitle: conda-hdmi2usb-packages/openocd at windows · timvideos/conda-hdmi2usb-packages · GitHub (at github.com)19:26
mithrohttps://github.com/mithro/conda-hdmi2usb-packages/blob/windows/openocd/bld.bat looks like I was using mingw ?19:27
tpbTitle: conda-hdmi2usb-packages/bld.bat at windows · mithro/conda-hdmi2usb-packages · GitHub (at github.com)19:27
FelixViyes, I currently don't have mingw19:27
mithroAs you can see it was a was quite a long time ago since I last worked on it...19:27
CarlFKconda complies things?19:28
FelixViI am trying to build those packages out of the anaconda prompt though19:28
FelixViusing "conda build xyz"19:28
mithroCarlFK[m]: Yes19:29
mithroCarlFK[m]: You can think of conda like dpkg19:29
CarlFKdpkg installs pre-compiled stuff19:29
mithroCarlFK[m]: But dpkg --build builds those packages19:29
CarlFKso that later someone else doen't have to19:30
FelixViin general, most packages have sh files but no bat files19:30
mithroCarlFK[m]: Yes, that is how conda works19:30
FelixViso I assume those were compiled under mingw then?19:30
mithroCarlFK[m]: We use travis to build conda packages so people don't have to compile stuff themselves19:30
CarlFKo does Felix need to build?19:30
CarlFKright.  so something is off.19:30
mithroCarlFK[m]: Because we haven't gotten that working for Windows19:30
CarlFKoh. thats whats off19:31
mithrohttps://conda.io/docs/user-guide/tasks/build-packages/compiler-tools.html -- https://anaconda.org/anaconda/mingw -- https://github.com/conda/conda-recipes/tree/master/gcc-519:34
tpbTitle: conda-recipes/gcc-5 at master · conda/conda-recipes · GitHub (at github.com)19:34
FelixViI tried to build openocd, installed patch, but it also fails at the end - much like flterm did19:35
FelixViit doesn't seem to find the binary19:35
tpbTitle: Tutorial: Basic tutorial for building a Conda package Conda documentation (at conda-test.pydata.org)19:35
mithroFelixVi: Do you have the output again?19:36
FelixViyes, trying to get it from the prompt into pastebin...19:43
FelixVigive me a sec19:43
tpbTitle: Possible Spam Detected (at pastebin.com)19:48
FelixVisorry about the lack of formatting - i'll have to figure out a better way to copy from that console19:48
mithroFelixVi: Can you run with --verbose?19:49
FelixVi--debug seems to be the closest thing19:52
tpbTitle: < CF-Cache-Status: HIT < CF-RAY: 3c8988ef1e2b4fe1-DEN < Date: Tue, 05 Dec 2017 - Pastebin.com (at pastebin.com)19:53
mithroI can't really see where it is doing the building....19:56
FelixViwhen I execute the run_test.bat manually, it just exits19:56
FelixViand whatever it does, the openocd binary is not found - which is why I think it exits in the first place19:57
FelixVithe subfolder in ..../conda-bld is empty19:59
FelixVithere's some subfolders but no files19:59
FelixViI assume the _b_env subfolder is where it would build19:59
tpbTitle: AppVeyor (at ci.appveyor.com)20:00
mithroIt looks like I didn't get as far as I thought i did...20:00
FelixViah, that would explain what's going on20:01
tpbTitle: Get packages working on Windows (including CI) · Issue #5 · timvideos/conda-hdmi2usb-packages · GitHub (at github.com)20:01
FelixViI'm interested in helping out on the Windows version20:03
FelixVibut it'll have to happen outside of work20:03
FelixViright now, I am trying to get some LM32 running to show that it's feasible at work20:04
FelixVido you think there is a way to get there relatively quick?20:04
mithroFelixVi: The quickest way would be a Linux VM?20:04
FelixViI got cygwin setup with gcc, binutils etc20:06
FelixVimisoc is running on it20:06
FelixVithat's why I started there - I think I got the environment up and running, too20:07
FelixVibut running make.py --platform mimasv2 is looking for hdmi2usb which doesn't exist20:07
FelixViI was just trying to build basesoc gateware for a simple test20:08
FelixVimaybe I should try to build for opsis first20:10
FelixVipython 3.6 doesn't work with the tracer in migen, but there's a workaround20:13
FelixVithis does the trick, but it's not backwards compatible: https://github.com/FelixVi/litex/commit/7df6c9110da7f43f8555539074d9314b68d4687d20:26
tpbTitle: Making tracer compatible with Python 3.6 · FelixVi/[email protected] · GitHub (at github.com)20:26
FelixVifor opsis, I now get to this point:20:28
FelixVimake: *** No rule to make target '#/HDMI2USB-litex-firmware/third_party/liteeth/firmware/uip/Makefile'.  Stop.20:28
FelixVibut there is no firmware folder that has been created20:29
*** thaytan has quit IRC20:41
*** thaytan has joined #timvideos20:42
*** ChanServ sets mode: +v thaytan20:42
mithroFelixVi: sorry, went to lunch21:21
FelixVimithro: ah, no worries21:21
FelixViCurrently, I'm trying to figure out the boot medium stuff21:22
FelixViif I pack bios into integrated rom, why can't I boot from flash through that bios?21:22
cr1901_modernrohitksingh: He has used it quite much.21:23
cr1901_modernNo, I have actively avoided it for the past 7 years. I happen to have it installed b/c I've been helping FelixVi out, and I've also needed it for the ATS programming language21:23
mithroFelixVi: What does "flashboot" show when you try?21:23
FelixViit's not listed under the "help" command21:24
FelixViso it says "command not found"21:24
FelixViwhen you have integrated-rom defined, flash doesn't get registered21:24
FelixVithat's what I find so confusing - I thought you can have bios in BRAM and use that to boot from flash21:25
FelixViwhen I build the basesoc without --integrated-rom-size I don't get any bios prompt anymore, so one seems to need that in order to get the bios into BRAM21:26
FelixVisee here: https://github.com/m-labs/misoc/blob/master/misoc/targets/papilio_pro.py#L8321:27
tpbTitle: misoc/papilio_pro.py at master · m-labs/misoc · GitHub (at github.com)21:27
FelixViI don't seem to be able to find the equivalent to this line here: https://github.com/timvideos/HDMI2USB-litex-firmware/blob/master/targets/mimasv2/base.py#L18621:30
tpbTitle: HDMI2USB-litex-firmware/base.py at master · timvideos/HDMI2USB-litex-firmware · GitHub (at github.com)21:30
mithroFelixVi: take a look at the Opsis + Atlys examples21:31
FelixVidoes opsis have the bios in BRAM or on flash?21:33
mithroFelixVi: All of the Pipistrello, Atlys and Opsis use BIOS in BRAM, Firmware in Flash21:34
mithroFelixVi: you will only get flashboot option if the bios knows you have SPI flash21:35
shorneshenki: switching back to uclibc I see the same issue again (after updating to the latest buildroot). It might be related to the PIC stuff (position independant code)21:39
shorneno... it cant be the PIC stuff, the variable sent to clock_gettime is not a global one21:46
*** sb0 has joined #timvideos22:00
FelixVimithro: are you supposed to get a bios prompt from BRAM when flash is found?22:18
mithroFelixVi: FYI - What do you have under HDMI2USB-litex-firmware/third_party/ directories?22:19
mithroFelixVi: Yes22:19
mithroFelixVi: What FPGA boards do you have again?22:20
FelixVipapilio pro, waxwing, saturn22:20
mithroThe waxwing + saturn should be easy to get going22:21
FelixViI got them setup, memtest is passing22:21
FelixViI just don't get the boot from flash part22:21
mithroCan you past the complete output?22:21
FelixViI can serial boot a test program I wrote22:21
mithroFelixVi: Oh -- you are much further along then I thought -- I assume you hadn't even seen anything from the BIOS22:22
FelixVionly with --integrated-rom-size22:23
FelixVias soon as I try to use flash, there is no more prompt22:23
mithroFelixVi: Can I see your soc config?22:24
tpbTitle: misoc/saturn.py at config · FelixVi/misoc · GitHub (at github.com)22:26
mithroFelixVi: you want something like this -> https://github.com/timvideos/HDMI2USB-litex-firmware/blob/master/targets/opsis/base.py#L261-L27122:28
tpbTitle: HDMI2USB-litex-firmware/base.py at master · timvideos/HDMI2USB-litex-firmware · GitHub (at github.com)22:28
mithroFelixVi: Remove the if here -> https://github.com/FelixVi/misoc/blob/config/misoc/targets/saturn.py#L13822:29
tpbTitle: misoc/saturn.py at config · FelixVi/misoc · GitHub (at github.com)22:29
mithroFelixVi: then add a "self.flash_boot_address" value22:29
FelixViI tried that, but since it's misoc it doesn't know constants22:30
FelixViwhat I have is hopefully as close as possible to your example22:30
cr1901_modern... hmmm, maybe you were right all along FelixVi re: it works in litex but not misoc22:31
mithroLet me look up what misoc currently uses22:32
FelixViah, thank you!22:32
FelixViI am not far enough along to fix this problem - I think it gets clearer once I see how things are supposed to work22:33
mithroFelixVi: the BIOS code is here -> https://github.com/m-labs/misoc/blob/master/misoc/software/bios/boot.c#L263-L28922:33
tpbTitle: misoc/boot.c at master · m-labs/misoc · GitHub (at github.com)22:33
FelixViI don't get the printf22:34
cr1901_modernNope, CONFIG_ constants are part of misoc, just checked22:34
FelixViit just sits silent22:34
tpbTitle: misoc/cpu_interface.py at master · m-labs/misoc · GitHub (at github.com)22:34
mithroIt does look like you should just be setting "self.flash_boot_address" value22:35
tpbTitle: misoc/builder.py at master · m-labs/misoc · GitHub (at github.com)22:35
FelixVito me it seems like the bios doesn't get executed when flash is specified22:35
mithroFelixVi: If you have that if statement, then it will try and use the bios from flash22:36
mithroFelixVi: Which isn't what you want22:36
FelixViah, that makes sense, too - if I remove it I get a wishbone collision22:36
FelixVithat's probably because mapping isn't right somewhere22:36
mithroFelixVi: Replace this line -> https://github.com/FelixVi/misoc/blob/config/misoc/targets/saturn.py#L14422:37
tpbTitle: misoc/saturn.py at config · FelixVi/misoc · GitHub (at github.com)22:37
mithro        self.register_mem("spiflash", self.mem_map["spiflash"], self.spiflash.bus, size=platform.spiflash_total_size)22:38
mithroSomething like that22:38
mithroFelixVi: BTW Did you actually flash both the gateware + bios?22:39
tpbTitle: Papilio pro default rom size is too small · Issue #60 · m-labs/misoc · GitHub (at github.com)22:39
FelixVithis is what I tried initially22:39
mithroFelixVi: Okay, it's most likely the spiflash isn't actually working correctly22:39
mithroFelixVi: Take a look at this file -> https://github.com/timvideos/HDMI2USB-litex-firmware/blob/master/targets/opsis/base.py#L261-L27122:40
tpbTitle: HDMI2USB-litex-firmware/base.py at master · timvideos/HDMI2USB-litex-firmware · GitHub (at github.com)22:40
cr1901_modernFelixVi: Wait a sec...22:40
cr1901_modernFelixVi: Nevermind. I just remembered something important, and that I suspect that spiflash1x does not current work properly in misoc.22:43
cr1901_modernspiflash2x should be fine22:43
cr1901_modernBut maybe it isn't. In any case, spiflash in litex _is_ known to work correctly22:44
cr1901_modern(This is why I disdain forks)22:44
cr1901_modernWell that's something I'll have to fix in misoc22:45
FelixVithe spi_flash in litex is very different22:46
FelixViI noticed that before, but didn't know what to make of it22:46
cr1901_modernCorrect, b/c _florent_ and I rewrote it after some issues w/ the misoc version22:46
FelixVibut in either event, shouldn't you still get some text via serial from the bios?22:47
FelixViI'm literally getting nothing22:47
cr1901_modernIdk :(22:47
FelixVieven if flash doesn't work at all, there's still some printf in the bios when that get's executed22:47
cr1901_modernIt's possible the bios is crashing before it reaches the text22:48
cr1901_modernI'd use a debug led for that22:48
tpbTitle: misoc/boot.c at master · m-labs/misoc · GitHub (at github.com)22:48
cr1901_modernbasically, misoc did not meaningfully support 1x spiflash. sb0 provided a fix but we opted to use our own impl for _reasons_22:48
FelixViah - but all the numato boards use 1x spiflash22:49
FelixViI assume that's why I am not having much luck22:49
FelixViIs there a simple way to use the mitex spiflash core in misoc?22:49
cr1901_modernFelixVi: Really what I should do is trim the litex core down and contribute it back to misoc22:50
FelixVicr1901_modern: do you have either a papilio, mimas, waxwing or saturn to try this with?22:50
cr1901_modernYes I have a mimasv222:50
FelixViah, that's why you said misoc didn't work but mitex did?22:50
cr1901_modernself.submodules.spiflash = spi_flash.SpiFlash(platform.request("spiflash2x"), dummy=8, div=4)- you're using a 2x here...22:51
FelixViisn't the flash the same as in papilio?22:51
cr1901_modern(5:47:04 PM) FelixVi: ah - but all the numato boards use 1x spiflash22:51
cr1901_modernI don't know- it may very well be.]22:51
FelixVisaturn uses m25p1622:52
FelixVipapilio pro uses mx25l644522:52
FelixViboth seem to be 1x to me22:53
FelixViso how did the papilio pro flash boot ever work?22:53
FelixViI'm just confused22:53
tpbTitle: PapilioPro (at papilio.cc)22:53
FelixViI have tried to get anything to boot on all 3 platforms thus far22:54
FelixVipipistrello is obsolete (or can you still get them)?22:55
FelixVimimas doesn't seem to work with misoc either (or does it?)22:55
FelixViI'd be more than happy to buy another board to get things started22:55
cr1901_modern... Let me get my bearings please ._.22:55
FelixViis there a S6 board that is known to work?22:55
cr1901_modernmimasv2, minispartan622:56
FelixVimimasv2 uses m25p16 - which is the same as waxwing and saturn22:56
cr1901_modernI can't even _find_ a datasheet for papilio's flash22:58
FelixViwhy don't we focus on mimasv2 then? since you got the board22:59
cr1901_modernFelixVi: Hold on I found it22:59
FelixVithat one is the same as the LX45 boards I got here22:59
FelixVipapilio was just what I tried since it was already part of misoc - I don't actually use those for anything except quick prototyping and testing22:59
cr1901_modernFelixVi: The Papilio Flash _does_ support 2x23:00
cr1901_modernSaturn does not23:00
FelixViwait, I thought there a bpi version and a non-bpi23:00
FelixVior is that different23:00
cr1901_modernI have no idea what that means23:00
*** sb0_ has joined #timvideos23:01
*** sb0 has quit IRC23:01
cr1901_modernFelixVi: Try w/ spiflash1x. Ignore what I said about "misoc being broken" for now; I want you to try anyway.23:04
cr1901_modern(try w/ misoc*)23:04
FelixVigive me one sec23:11
FelixVioh, just need to modify the platform23:13
FelixViwhat pin becomes dq?23:14
cr1901_modernRight, spiflash1x and spiflash2x have the same connections23:14
FelixViin spiflash 2x, dq is mosi, miso23:14
cr1901_modernmiso should be dq; mosi never switches directions in 1x23:15
FelixViso wouldn't 1x only have 1 dq pin?23:15
cr1901_modernactually it shouldn't be called dq.23:15
* cr1901_modern checks23:15
tpbTitle: ("spiflash", 0, Subsignal("cs_n", Pins("V3")), Subsignal(" - Pastebin.com (at pastebin.com)23:16
FelixVithis is my 1x flash interface23:16
FelixViand misoc tells me "AttributeError: 'Record' object has no attribute 'dq'" when I try to use that23:16
FelixViso it seems like it wants dq23:17
mithroFelixVi: We have the MimasV2 working on the HDMI2USB-litex-firmware23:17
cr1901_modernFelixVi: I retract my retraction. Use Litex until I figure out what to do.23:18
mithroFelixVi: Mimas != MimasV223:18
mithroFelixVi: https://github.com/timvideos/HDMI2USB-firmware-prebuilt/blob/master/archive/master/v0.0.3-1058-g0c06257/mimasv2/base/lm32/image-gateware%2Bbios%2Bfirmware.bin23:18
tpbTitle: HDMI2USB-firmware-prebuilt/image-gateware+bios+firmware.bin at master · timvideos/HDMI2USB-firmware-prebuilt · GitHub (at github.com)23:18
FelixViand mimas has the same flash as saturn and waxwing23:18
mithroFelixVi: Give that image a go23:18
FelixVimithro: I didn't get litex up and running yet23:18
mithroFelixVi: If you flash that above .bin you can see if it works on your board23:19
FelixVioh, are those going to work on saturn?23:19
mithroFelixVi: No23:19
FelixVimithro: I don't have a mimasv223:19
cr1901_modernFelixVi doesn't have mimasv223:19
mithroOh - why are we talking about a MimasV2?23:19
FelixVibut I'm happy to order one23:19
FelixVimimasv2 and saturn, waxwing have the same spi flash23:19
cr1901_modernmithro: B/c spiflash boot is known to work23:20
mithroFelixVi: Did you make the changes I said to your saturn soc?23:20
FelixViyes, that's the config I posted23:21
FelixVibut misoc and litex are different when it comes to spi flash23:21
FelixViso I could only get it so close23:21
mithroFelixVi: I don't see any changes in https://github.com/FelixVi/misoc/blob/config/misoc/targets/saturn.py#L144 ?23:21
tpbTitle: misoc/saturn.py at config · FelixVi/misoc · GitHub (at github.com)23:21
cr1901_modernFelixVi: You have managed to stumble upon pretty much every ugly wart in migen/misoc and litex development in two weeks by trying to port a new board to _misoc_ (em).23:22
cr1901_modernClearly something we need to work on.23:22
FelixVimithro: this is what I currently have23:23
tpbTitle: misoc/saturn.py at config · FelixVi/misoc · GitHub (at github.com)23:23
cr1901_modernFelixVi: Litex/MiSoC diverged in that respect, and IMO they never should have. But I just didn't have the energy to make them converge again.23:23
FelixVicr1901_modern: Sorry, that wasn't my intention ;)23:23
cr1901_modernBut since an actual user ran into this issue I don't think I have a choice now23:23
FelixViI'll help where I can23:23
FelixViAt this point, it seems like I'm not crazy and it's clear that I wasn't doing things in a weird way23:24
FelixViSo I might try to move beyond trying to figure out what I'm doing wrong to fixing the issue23:24
FelixViBut neither FHDL nor spi flash are areas I have a lot of experience in23:25
cr1901_modernFelixVi: Could you make a repo/gist of your saturn migen config and misoc impl that bombs with "dq" is required?23:25
tpbTitle: misoc/saturn.py at config · FelixVi/misoc · GitHub (at github.com)23:25
mithroFelixVi: So what happens when you use that code? I think that would just explode with a traceback?23:25
FelixVimithro: yeah, I get an index error23:26
cr1901_modernFelixVi: The issue is that the naming convention for spiflash1x doesn't match spiflash2x and above23:26
mithroFelixVi: Ignore cr1901_modern for now23:26
tpbTitle: $ python3 -m misoc.targets.saturn fvco (MHz) = 1000 n= 10 d= 1 clk_freq (par - Pastebin.com (at pastebin.com)23:26
mithroFelixVi: I assume you get a key error about spiflash not existing in mem_map?23:26
mithroFelixVi: You need to add these lines -> https://github.com/timvideos/HDMI2USB-litex-firmware/blob/master/targets/opsis/base.py#L231-L23423:26
tpbTitle: HDMI2USB-litex-firmware/base.py at master · timvideos/HDMI2USB-litex-firmware · GitHub (at github.com)23:26
FelixViit crashes here: https://github.com/FelixVi/misoc/blob/config/misoc/targets/saturn.py#L13923:27
tpbTitle: misoc/saturn.py at config · FelixVi/misoc · GitHub (at github.com)23:27
FelixVimem_map is where it would crash after the current issue23:28
mithroFelixVi: Traceback?23:28
FelixViin the pastebin above23:28
FelixVicrashes on " self.miso.status.eq(dq.i[1]) "23:29
FelixVibecause the 1x intf has inconsistent pin naming conventions compared to 2x and 4x23:29
FelixVibut - why does my papilio pro also not work in 2x mode?23:31
FelixVicr1901_modern: are you in the us? I can mail you a papilio board if that helps move things along23:32
mithroFelixVi: add "with_bitbang=False"23:32
cr1901_modernFelixVi: Yes, but that won't be necessary23:32
FelixVimithro: now we're on the spiflash_total_size error - I don't have that defined in my platform23:33
mithroFelixVi: https://github.com/timvideos/HDMI2USB-litex-firmware/blob/master/platforms/mimasv2.py#L132-L14323:34
tpbTitle: HDMI2USB-litex-firmware/mimasv2.py at master · timvideos/HDMI2USB-litex-firmware · GitHub (at github.com)23:34
FelixViTypeError: register_mem() got an unexpected keyword argument 'size'23:35
FelixVilet me get you a snapshot of my platform, too23:35
mithroFelixVi: Remove ", size=platform.spiflash_total_size" from the register_mem line23:36
FelixViTypeError: register_mem() missing 1 required positional argument: 'interface'23:36
mithroFelixVi: register_mem(self, name, origin, length, interface):23:38
FelixViself.register_mem("spiflash", self.mem_map["spiflash"], platform.spiflash_total_size, self.spiflash.bus)23:39
mithroself.register_mem("spiflash", self.mem_map["spiflash"], platform.spiflash_total_size, self.spiflash.bus)23:39
mithrojinx :-P23:39
FelixViok, we had the same idea23:39
FelixVimigen.fhdl.module.FinalizeError: CPU needs a rom to be registered with register_mem()23:39
cr1901_modernmithro: Am I allowed to speak now?23:39
mithroFelixVi: Are you still giving "integrated_rom = False" to the SoC somewhere?23:40
FelixViI just changed it to 2000023:40
FelixViit now synthesizes23:40
mithrocr1901_modern: As soon as FelixVi gets a bios which boots with the "flashboot" command you can take over getting the spiflash to actually work23:40
cr1901_modernmithro: Fine. In any case I added it to my internal "misoc/litex" todo list since this appears to be a legitimate bug and not user error23:41
FelixViI am not sure if the bitfile offsets are right - but we'll see23:41
mithroFelixVi: Your "        self.flash_boot_address = 0x180000" is wrong23:42
FelixViok, then upload top to 0x0 and firmware to 0x180000?23:42
mithroFelixVi: The Saturn has a LX45 right?23:43
FelixViyeah, so the bitfile is ~5x larger than for LX923:43
tpbTitle: HDMI2USB-litex-firmware/atlys.py at master · timvideos/HDMI2USB-litex-firmware · GitHub (at github.com)23:43
mithroFelixVi: You'll want these lines too23:44
tpbTitle: HDMI2USB-litex-firmware/atlys.py at master · timvideos/HDMI2USB-litex-firmware · GitHub (at github.com)23:44
FelixVitop.bit is 1,485,325 bytes large23:44
FelixVioh, that's pretty much the same size23:44
mithroThe Atlys is a LX4523:44
FelixVioh, that makes a lot of sense - thought we're looking at mimasv223:45
FelixVik, got that in there23:46
mithroFelixVi: It would have been better to start from the Atlys or the Pipistrello if I had realized that you were also using a LX4523:47
mithroFelixVi: Upload your latest and I'll take a look23:47
FelixVithis is the migen part: https://github.com/FelixVi/migen/blob/configuration/migen/build/platforms/saturn.py23:49
tpbTitle: migen/saturn.py at configuration · FelixVi/migen · GitHub (at github.com)23:49
FelixViand this is misoc: https://github.com/FelixVi/misoc/blob/config/misoc/targets/saturn.py23:49
tpbTitle: misoc/saturn.py at config · FelixVi/misoc · GitHub (at github.com)23:49
mithroFelixVi: This still needs fixing -> https://github.com/FelixVi/misoc/blob/config/misoc/targets/saturn.py#L14823:50
tpbTitle: misoc/saturn.py at config · FelixVi/misoc · GitHub (at github.com)23:50
FelixViself.flash_boot_address = self.mem_map["spiflash"]+platform.gateware_size+bios_size23:51
FelixViand bios-size is --integrated-rom-size, correct?23:51
mithroFelixVi: Yes23:53
FelixVik, just pushed that again23:53
mithroFelixVi: remove this line -> https://github.com/FelixVi/misoc/blob/config/misoc/targets/saturn.py#L12423:53
tpbTitle: misoc/saturn.py at config · FelixVi/misoc · GitHub (at github.com)23:53
FelixVik, got that23:54
FelixViI'm not gonna ask about what that does now, hopefully I can figure it out on my own23:55
mithroFelixVi: It sets the address that the CPU starts executing at :-P23:57
FelixViso, 0 is the default which we want now23:59
FelixViis 0-20000 the bram bios in our memory map?23:59

Generated by irclog2html.py 2.13.1 by Marius Gedminas - find it at mg.pov.lt!